Conventional data management systems that target high-availability have to allow transactional operations, such as updates, deletes, and inserts (UDIs) to go to any replica of data. The transactional operations also target compatibility with the big data ecosystem, which uses append-only (and hence mutation unfriendly) storage streams because of their superiority in efficient read and write operations and space consumption.